Thematic CERN School of Computing on Scientific Software for Heterogeneous Architectures 2025

Name: Thematic CERN School of Computing on Scientific Software for Heterogeneous Architectures 2025
Start: 2025-10-05T15:00:00+02:00
End: 2025-10-11T11:30:00+02:00
Location: Medils Institute, Split

5 Oct 2025, 15:00 → 11 Oct 2025, 11:30 Europe/Zagreb

Medils Institute, Split

Meštrovićevo šetalište 45 HR – 21000 Split Croatia

Alberto Pace (CERN), Kristina Gunne (CERN), Andrzej Nowicki (CERN)

Description

The 18^th Thematic CERN School of Computing (tCSC 2025) will take place on October 5-11, 2025. The theme of the school is "Scientific Software for Heterogeneous Architectures" - see the academic programme for more details.

This school is organized by CERN in collaboration with the Faculty of Science, University of Split. The school will take place in Split, Croatia, and be hosted at the Mediterranean Institute for Life Sciences (MEDILS) Conference Centre. The Centre is a historical renovated building situated in a wooded and landscaped park located on the Adriatic Sea coast, a few kilometers from the center of Split.

Target Audience

The School is aimed at postgraduate (i.e. a minimum of Bachelor's degree or equivalent) students, engineers, and scientists with a few years of experience in particle physics, in computing, or in related fields. We welcome applications from all countries and nationalities.

Important dates

14 May application opens
10 July application close
23 July invitations sent to selected students
20 August confirmation of participation deadline
3 September participation fee deadline

CERN School of Computing

Computing.School@cern.ch

Monday 6 October
- Mon 6 Oct
- Tue 7 Oct
- Wed 8 Oct
- Thu 9 Oct
- Fri 10 Oct
- Sat 11 Oct
- 09:00 → 10:00
  
  Opening Session 1h CERN
  
  CERN
- 10:00 → 10:30
  
  Self-presentation: 1 minute per person 30m CERN
  
  CERN
  
  self_presentations.pdf
- 10:30 → 11:00
  
  Coffee Break 30m CERN
  
  CERN
- 11:00 → 12:30
  
  Lecture 1 1h 30m CERN
  
  CERN
  
  Description of theme / lecture
- 12:30 → 13:30
  
  Lunch 1h
- 13:30 → 15:00
  
  Practice 1 1h 30m CERN
  
  CERN
  
  Description about Practice 1
- 15:00 → 15:30
  
  Coffee Break 30m CERN
  
  CERN
- 15:30 → 17:00
  
  Practice 2 1h 30m CERN
  
  CERN
  
  Description about Practice 1
Tuesday 7 October
- Mon 6 Oct
- Tue 7 Oct
- Wed 8 Oct
- Thu 9 Oct
- Fri 10 Oct
- Sat 11 Oct
- 07:45 → 08:45
  
  Breakfast 1h
- 08:45 → 09:45
  Writing parallel software 1h
  Parallel and Optimised Scientific Software - lecture 1
  - Amdahl's and Gustafson's laws
  - Asynchronous execution
  - Finding concurrency, task vs. data parallelism
  - Using threading in C++ and Python, comparison with multi-process
  - Resource protection and thread safety
  - Locks, thread local storage, atomic operations
  Speaker: Sebastien Ponce (CERN)
  
  WritingParallelSoft-pres.pdf
- 09:45 → 10:45
  Writing efficient software 1h
  Parallel and Optimised Scientific Software - lecture 2
  - Why Python and C++ ?
  - Recent evolutions: C++ 11/14/17
  - Modern features of C++ related to performance
  - Templating versus inheritance, pros and cons of virtual inheritance
  - Python 3, and switching from Python 2
  Speaker: Sebastien Ponce (CERN)
  
  WritingEfficientSoft-pres.pdf
- 10:45 → 11:00
  
  Announcements 15m
- 11:00 → 11:05
  
  School Photo 5m
- 11:05 → 11:30
  
  Coffee break 25m
- 11:30 → 12:30
  Low-level Performance Optimization 1h
  CPU Architecture and High Performance - lecture 3
  - Main sources of performance bottlenecks
  - Floating point arithmetics performance
  - Advanced low-level performance tuning
  Speaker: Guilherme Amadio (CERN)
  
  03.pdf
- 12:30 → 13:30
  
  Data-Oriented Design 1h
  
  CPU Architecture and High Performance - lecture 4
  - Principles of data-oriented design
  - Memory access and data-type profiling
  - Data structure performance optimization
  
  Speaker: Guilherme Amadio (CERN)
  
  04.pdf
- 13:30 → 14:30
  
  Lunch 1h
- 14:30 → 16:00
  
  Study or sports time 1h 30m
- 16:00 → 16:30
  
  Coffee break 30m
- 16:30 → 17:30
  Optimizing existing large codebase 1h
  Parallel and Optimised Scientific Software - lecture 3
  - Measuring performance, tools and key indicators
  - Improving memory handling
  - The nightmare of thread safety
  - Code modernization and low level optimizations
  - Data structures for efficient computation in modern C++
  Speaker: Sebastien Ponce (CERN)
  
  LargeCodeOptimization-pres.pdf
- 17:30 → 18:15
  Student lightning talks 45m
  - Studies of a new track fitting algorithm for the ATLAS Event Filter 7m
    
    The ATLAS experiment is going through a comprehensive set of upgrades in preparation for data taking at the High-Luminosity Large Hadron Collider. The ATLAS Event Filter (EF) provides a second level of filtering, reducing the Level-0 trigger rate of 1 MHz to 10 kHz for permanent storage. To this end, it is required to perform track reconstruction (EF Tracking) for the entire Inner Tracker (ITk) at a maximum rate of 150 kHz. A new parallelizable track fit, based on hit triplets, has been developed and is being studied for use in EF Tracking. The General Triplet Track Fit (GTTF) [10.1016/j.nima.2025.170391] is a generalization of the Multiple Scattering-only triplet fit [10.1016/j.nima.2016.11.012], developed originally for the Mu3e experiment, and includes hit uncertainties. The status of ongoing studies of the GTTF for use in EF Tracking will be presented.
    
    Speaker: Abhirikshma Nandi (Heidelberg University (DE))
    
    tCSC Lightening Talk.pdf
  - Using GPUs at HPC centers for CMS 7m
    
    Overview about how GPUs at the National High-Performance Computing Center (HoreKa) in Karlsruhe can be used for CMS workflows and by CMS users.
    
    Speaker: Nikita Shadskiy (KIT)
    
    2025_10_07_tCSC25.pdf
  - Quantum-Centric Supercomputing: Bridging Quantum & Classical Worlds 7m
    
    Quantum-centric supercomputing is an approach that integrates quantum hardware with classical high-performance computing to solve complex problems more efficiently. Rather than placing quantum computers in isolation, this paradigm enables them to work collaboratively with classical systems; allocating tasks to whichever component is best suited for each step
    
    Speaker: Alexander Varga (University of Calgary)
    
    CERN -Lightning Talk.pdf
    
    CERN -Lightning Talk.pptx
  - A Practical Introduction to Microbenchmarking 7m
    
    Once a profiler identifies a performance bottleneck, the subsequent challenge is to diagnose its root cause. This talk introduces microbenchmarking as a methodology for this purpose. Using google-benchmark, we will demonstrate how to measure granular code segments to establish performance limits, quantify overhead, and analyze hardware-software interactions, providing a systematic approach to code optimization.
    
    Speaker: Michael Lettrich (CERN)
    
    Microbenchmarking.pdf
  - Finally, A Good Use For AI! 7m
    
    Speaker: Robin Hofsaess
    
    Finally, I found a good use for AI!_.pptx
- 18:15 → 19:15
  
  CPU Architecture and High Performance - exercise 2 1h
  
  Speaker: Guilherme Amadio (CERN)
  
  exercise2.pdf
  
  Further Reading: Advanced Matrix Multiplication Optimization on Modern Multi-Core Processors
  
  Further Reading: Optimizing matrix multiplication
  
  LLVM Sample: Matrix Multiplication (what this exercise is based on)
  
  matrix.c
  
  VTune Sample: Matrix Multiplication
- 19:30 → 20:30
  
  Dinner 1h
Wednesday 8 October
- Mon 6 Oct
- Tue 7 Oct
- Wed 8 Oct
- Thu 9 Oct
- Fri 10 Oct
- Sat 11 Oct
- 07:45 → 08:45
  
  Breakfast 1h
- 08:45 → 09:45
  Scientific computing on heterogeneous architectures 1h
  Programming for Heterogeneous Architectures - lecture 1
  - Introduction to heterogeneous architectures and the performance challenge
  - From general to specialized: Hardware accelerators and applications
  - Type of workloads ideal for different accelerators
  - Trade-offs between multi-core and many-core architectures
  - Implications of heterogeneous hardware on the design and architecture of scientific software
  - Embarrassingly parallel scientific applications in HPC and CERN
  Speaker: Daniel Hugo Campora Perez (NVIDIA)
  
  0_scientific_computing_on_heterogeneous_architectures.pdf
- 09:45 → 10:45
  
  Optimizing existing large codebase - exercise 1 1h
  
  Optimisation of an existing, production grade large codebase
  
  Speaker: Sebastien Ponce (CERN)
  
  LargeCodeExercise.pdf
- 10:45 → 11:00
  
  Announcements 15m
- 11:00 → 11:15
  
  Coffee break 15m
- 11:15 → 12:15
  
  Optimizing existing large codebase - exercise 2 1h
  
  Optimisation of an existing, production grade large codebase
  
  Speaker: Sebastien Ponce (CERN)
- 12:15 → 13:00
  
  Lunch 45m
- 13:00 → 18:00
  
  River rafting excursion - Cetina River 5h
- 18:30 → 20:30
  
  Dinner at Kastel Slanica Omis 2h
- 20:30 → 21:30
  
  Transport to Medils 1h
Thursday 9 October
- Mon 6 Oct
- Tue 7 Oct
- Wed 8 Oct
- Thu 9 Oct
- Fri 10 Oct
- Sat 11 Oct
- 07:45 → 08:45
  
  Breakfast 1h
- 08:45 → 09:45
  Performant programming for GPUs 1h
  Programming for Heterogeneous Architectures - lecture 2
  - From SIMD to SPMD, a programming model transition
  - Thread and memory organization
  - Basic building blocks of a GPU program
  - Control flow, synchronization, atomics
  Speaker: Daniel Campora (NVIDIA)
  
  1_programming_for_gpus.pdf
- 09:45 → 10:45
  Performant programming for GPUs 1h
  Programming for Heterogeneous Architectures - lecture 3
  - Data locality, coalesced memory accesses, tiled data processing
  - GPU streams, pipelined memory transfers
  - Under the hood: branchless, warps, masked execution
  - Debugging and profiling a GPU application
  Speaker: Daniel Campora (NVIDIA)
  
  2_performant_programming_for_gpus.pdf
- 10:45 → 11:00
  
  Announcements 15m
- 11:00 → 11:30
  
  Coffee break 30m
- 11:30 → 12:30
  Practical vectorization 1h
  Parallel and Optimised Scientific Software - lecture 4
  - Measuring vectorization level
  - What to expect from vectorization
  - Preparing code for vectorization
  - Vectorizing techniques in C++: intrinsics, libraries, autovectorization
  Speaker: Sebastien Ponce (CERN)
  
  PracticalVectorization-pres.pdf
- 12:30 → 13:30
  
  Practical vectorization - exercise 1h
  
  Speaker: Sebastien Ponce (CERN)
  
  VectorizationExercise.pdf
- 13:30 → 14:30
  
  Lunch 1h
- 14:30 → 16:30
  
  Study time or sports 2h
- 16:30 → 17:00
  
  Coffee break 30m
- 17:00 → 18:00
  
  Programming for heterogeneous architectures - exercise 1h
  
  Speaker: Daniel Campora (NVIDIA)
  
  instructions_to_start_gpu_lab.pdf
- 18:00 → 19:00
  
  Programming for heterogeneous architectures - exercise 1h
  
  Speaker: Daniel Campora (NVIDIA)
- 19:30 → 20:30
  
  Dinner 1h
- 20:30 → 21:30
  
  The Future of the Universe and Humanity 1h
  
  Speaker: Ivica Puljak (Technical University of Split FESB)
Friday 10 October
- Mon 6 Oct
- Tue 7 Oct
- Wed 8 Oct
- Thu 9 Oct
- Fri 10 Oct
- Sat 11 Oct
- 07:45 → 08:45
  
  Breakfast 1h
- 08:45 → 09:45
  
  Debrief of exercises 1h
- 09:45 → 10:45
  Design patterns and best practices 1h
  Programming for Heterogeneous Architectures - lecture 4
  - Good practices: single precision, floating point rounding, avoid register spilling, prefer single source
  - Other standards: SYCL, HIP, OpenCL
  - Middleware libraries and cross-architecture compatibility
  - Reusable parallel design patterns with real-life applications
  Speaker: Daniel Campora (NVIDIA)
  
  3_design_patterns_and_best_practices.pdf
- 10:45 → 11:00
  
  Announcements 15m
- 11:00 → 11:30
  
  Coffee break 30m
- 11:30 → 12:30
  
  Programming for heterogeneous architectures - exercise 1h
  
  Speaker: Daniel Campora (NVIDIA)
  
  instructions_to_start_gpu_lab.pdf
- 12:30 → 13:30
  
  Programming for heterogeneous architectures - exercise 1h
  
  Speaker: Daniel Campora (NVIDIA)
- 13:30 → 14:30
  
  Lunch 1h
- 14:30 → 15:30
  
  Exam 1h
- 15:30 → 16:00
  
  Coffee break 30m
- 16:00 → 18:30
  
  Free time 2h 30m
- 18:30 → 19:30
  
  Closing Session 1h
- 19:30 → 21:30
  
  Closing dinner Medils 2h
Saturday 11 October
- Mon 6 Oct
- Tue 7 Oct
- Wed 8 Oct
- Thu 9 Oct
- Fri 10 Oct
- Sat 11 Oct
- 07:30 → 08:30
  
  Breakfast 1h
- 09:00 → 11:30
  
  Departures 2h 30m

Choose timezone

Thematic CERN School of Computing on Scientific Software for Heterogeneous Architectures 2025

Medils Institute, Split

CERN

CERN

CERN

CERN

CERN

CERN

CERN